The purpose of this paper is to analyze the main types of core literacy of sports professionals in colleges and universities using association rules and the XGBoost algorithm. Combined with the confidence rule in association analysis, it finds out the minimum support degree of all the core qualities that satisfy the core qualities of sports professionals in colleges and universities. The efficiency of modeling talent demand can be improved by using the Apriori algorithm. Adding new regression trees sequentially based on the original model using the XGBoost algorithm. The college sports talent demand model was constructed by combining association rules and the XGBoost algorithm. Based on the results, composite talents, professional talents, and application talents are the main types of sports professional talent cultivation positioning. The cultivation ratio of sports composite talents in University A is 0.7, and the cultivation ratio of application composite talents is 0.6. The school’s requirements for sports talent’s ability are the highest, resulting in a cultivation ratio of 0.75 for sports talent. This study promotes, to a certain extent, the cultivation of sports professionals in colleges and universities.